Who are the publics engaging in AI?

Area of Science:

• Artificial Intelligence
• Public Engagement
• Science and Technology Studies

Background:

• Public engagement is crucial for governmental adoption of artificial intelligence (AI) systems.
• AI researchers and practitioners often neglect to define or categorize the publics involved in AI development and deployment.
• The definition of 'publics' influences assumptions about their capacity for policy contribution and knowledge production in AI.

Purpose of the Study:

• To analyze how 'publics' are defined and framed within the scientific literature on artificial intelligence.
• To explore the implications of different public classifications on AI policy and knowledge co-production.
• To identify the semantic nuances, affordances, and political/expertise lenses used to describe publics in AI.

Main Methods:

• A structured analysis employing a mixed-method approach.
• Algorithmic generation of search terms to initially examine approximately 2500 articles.
• Systematic literature review of approximately 100 selected documents.

Main Results:

• The study identified a multiplicity of ways 'publics' are framed in AI literature.
• Analysis revealed diverse semantic nuances, affordances, and varying political and expertise lenses applied to publics.
• A significant lack of clear, consistent definitions for 'publics' was observed across the reviewed literature.

Conclusions:

• Categorizing publics in AI is not neutral; it involves acts of power, politics, and truth-seeking.
• The framing of publics shapes their perceived role and influence in AI governance and development.
• Further research is needed to establish clearer, more inclusive definitions of publics in the context of AI.